Transition from Rork while two delivery systems share the road

Start with an inventory and a thin operational baseline, then move one boundary at a time: observability, read models, authentication, writes, and finally release control. Each stage has a measured exit condition and a short overlap window. We prefer a compatibility API or event bridge over copying screens, because it lets the product team prove behavior while Rork still carries a known-good workload.

Dual-track delivery requires one authority for each record, not two systems both claiming to be correct. We route reads deliberately, make writes idempotent, reconcile deltas, and expose a kill switch for the new path. Background jobs and notifications need the same ownership rules. “Both are live” is safe only when duplicate sends and conflicting updates are impossible or recoverable.

The organization—not an individual builder account—should own Apple and Google developer access, bundle/package identifiers, certificates, signing keys, privacy declarations, and reviewer contact details. We create a release checklist with named approvers and backup access before the first staged binary. The transition is not complete if the app moved but the ability to submit an emergency fix did not.

Plan the first live week around US Central business hours with an explicit escalation owner, release freeze windows, dashboard links, and customer-support language. We front-load rehearsals and keep the old path available long enough to make a controlled retreat. Coverage means someone can interpret a failed job or auth spike, not merely someone being reachable in chat.